Economic Price Optimization Part 5: Useful, Not Perfect

May 2013 Pricing 1 Comment

From the time we were kids, we have heard about economics and understood that it deals with prices.  Now that we are grown-ups with jobs, it seems like using an economic tool would be the right thing to do when it comes to pricing.  But, as grown-ups, we also know that things are never as simple as they appeared when we were kids.

We are executives.  We manage complexity under uncertainty with limited resources and time towards goals which enable our firms to thrive and our customers to reach satisfaction.  We know there are many tools we can throw at a challenge in the aim of reaching our goals.  We also know that simply throwing tools at a problem on an ad hoc and arbitrary basis to see if they work is a waste of time and resources.

By the same logic, we know that proper pricing isn’t simply a matter of applying a singly defined economic price optimization tool at our pricing questions.  Rather, we seek to identify the optimal prices for our offerings.  In terms of choosing the technique for reaching that goal, we select the one which provides the best tradeoff in accuracy and efficiency.

Alternatives in the Simple

As we have shown in this series, the term “economic price optimization” is an umbrella term for various techniques, not one single tool.  We have shown that economic price optimization can refer to simple theoretical studies of market behavior using linear demand curves.  We have shown that economic price optimization can refer to elasticity and cross-product measurements.  We’ve shown that that economic price optimization can refer to simple A/B market tests that identify which of two prices is better.  And across all of these meanings of economic price optimization, we have demonstrated that economic price optimization is highly dependent on understanding the real demand curve facing the business, and that fully defining that demand curve is far from a simple task.

A Complex Alternative: Data Mining Pricing

There is one other major meaning for the term “economic price optimization” that is worth mentioning, though calling it economic price optimization is stretch of the term.  Let’s just call it data mining pricing.

In data mining pricing, executives accept that they can’t optimize each individual price across their thousands of goods, or across each of their thousands of customers.  But even if they can’t optimize each individual price, they can segment their customers and products, and at least “optimize” the price within that segment.  From this process, some of the individual prices within the segment may be too low while others may be too high, but overall, they are better prices than would have arrived at otherwise.  By better, I mean that customers find the prices to be reasonable and purchase, and that the firm finds that these prices improve its profits.

Data mining pricing doesn’t have to be big.  For some challenges, executives should keep it simple with off-the-shelf statistical software and a good data analyst.  For others, dedicated software from firms like Vendavo, Zilliant, Pros, Vistaar, Stratinis, or SAP may be more appropriate.

Better is Better

Moreover, if you have been paying close attention to this series, you know that the price optimized at one point in time may not be the right price for another point in time.  The data mining pricing processes take that into account.  It depends on human beings reviewing the individual prices coming out of a pricing data mining exercise to ensure the output is reasonable.  This process also depends on human beings checking after the fact to ensure that the goods that were re-priced higher are still being sold, and that the goods that were re-priced lower aren’t leading the company into an unintended price war.   It requires iterative improvements.

Data mining pricing also accepts that the issue in pricing isn’t to find the perfect price every time but rather to identify a better price than it would have identified otherwise.  By better, here we mean one that enables the firm to outperform its competitors.  To outperform, you don’t have to be perfect, you just have to beat the competition more often than it beats you.

In analogy, the best stock picker doesn’t have to pick the right stock 100% of the time, they only need to pick it 51% of the time, and they will beat the market.  Similarly, some leading executives of pricing have said that the best big data pricing process doesn’t have to identify a better price 100% of the time, only 51% of the time, for it to be worth the effort.

Data mining pricing is a compromise and requires tradeoffs.  I know some of you don’t like compromises and tradeoffs, but grown-ups know that compromises are better than analysis paralysis.  And in truth, economics isn’t about pricing, it is about how people achieve their goals with scarce resources through compromises and tradeoffs.   At the end of the day, measurable improvement now beats out an intangible and fleeting ‘ideal price’ achieved in the rear view mirror.


About the author

Tim J. Smith, PhD is the Managing Principal of Wiglaf Pricing, and an Adjunct Professor at DePaul University of Marketing and Economics. His most recent book is Pricing Strategy: Setting Price Levels, Managing Price Discounts, & Establishing Price Structures.